1. Field of the Invention
The present invention relates to methods of fabricating semiconductor devices. More specifically, embodiments of the present invention provide an interconnect assembly and Z-connection method for coupling together fine pitch substrates.
2. Description of the Prior Art
Multi-chip modules (MCM) are substrates on which more than one integrated circuit chip is mounted. The substrate (e.g., a chip carrier) typically includes bonding pads for the chip, where the pads may be connected by a set of conductive lines to contact pads for a package or to other bonding pads on the substrate. The conductive lines thus form an interconnect network or I/O path for multiple elements mounted on a common substrate. Although there are other methods for electrically connecting an integrated circuit chip or other component to a substrate, the most commonly used ones are termed flip-chip bonding, wire bonding, and tape automated bonding (TAB).
In flip-chip (or solder bump) bonding, solder bumps are placed on the I/O pads of a chip and reflowed to form a bond with the chip pads. The chip is placed on a holder in a face-up position, flipped over (i.e., face-down) and aligned with the corresponding conductive bonding pads on the substrate. The chip pads and substrate pads are then brought into physical contact. The solder is reflowed by application of heat, causing the bumps to fuse with the bonding pads and provide both an electrical and structural connection between the chip""s I/O pads and the substrate. In wire bonding processes, the chips are attached to the substrate in a face-up position and thin gold or aluminum wires are then connected between the I/O pads of the chip and the bonding pads of the substrate. The wires are connected to the two sets of pads by means of a thermocompresison, thermosonic, or ultrasonic welding operation. In TAB bonding, the chips are attached to the substrate in a face-up position and the I/O pads of the chips are then bonded to metal pads on a polyimide film tape by either reflowing or thermocompression/ultrasonic bonding metal and/or solder bumps placed on the tape.
Each of the described methods for interconnecting chips to a substrate has its associated advantages and disadvantages. Flip-chip methods provide the highest density of I/O interconnects and therefore the capability of producing the smallest MCMs. The I/O connections may be made at the periphery of the chip or in its interior. The flip-chip bonds provide a good electrical connection, but the solder joints generally exhibit poor heat dissipation capability (low thermal conductivity). Another disadvantage is that the integrity of the solder joint between the pads can be reduced by thermally induced metal fatigue (differential thermal expansion between the components), and by corrosion caused by trapped solder flux or contaminants. Shorting between closely spaced bumps can occur when the solder is reflowed. In addition, because the solder must be heated to a sufficient temperature to cause it to reflow, flip-chip bonding may not be suited for use with some temperature sensitive components (i.e., components which are damaged, or have their electrical characteristics unacceptably altered when heated to solder reflow temperatures).
Wire bonding is a mature technology, however, the wires used in wire bonding are purposefully made very thin, thereby limiting the power they can transfer without failure. In addition, the lead inductance and resistance of the wires result in a degradation of the electrical performance of the interconnects. Wire bond connections have a larger footprint than flip-chip interconnects, and thus require a comparatively larger substrate than flip-chip bonded MCMs. The wires also form a relatively long signal propagation path compared to other interconnect methods, In addition, embrittlement of the interconnections as a result of the formation of intermetallics can cause a failure of the bonds (e.g., this is a problem between gold wirebonds and aluminum pads).
TAB interconnections offer the benefits of a smaller bonding pad and pitch than wire bonding. However, like wire bonding, TAB is limited to the interconnection of chips having perimeter I/O pads. This typically results in a lower overall I/O density than can be obtained using flip-chip technologies. TAB interconnects also generally have a higher capacitance and greater parasitic inductance than do flip-chip bonds. Finally, because TAB assembly usually requires different tooling for each chip design, TAB is relatively expensive bonding method. Therefore, what is desired and what has been invented is a method for interconnecting integrated circuit chips to a substrate which overcomes some of the disadvantages of presently available methods, while retaining many of the benefits of those methods. It is particularly desired to have a method for interconnecting chips to a substrate which has the benefits of flip-chip bonding, without the drawbacks of that method.
The present invention provides a fluxless method for forming an interconnect assembly comprising a first semiconductor substrate having a first pad connected thereto, and a post connected to the first pad and having a length greater than a thickness of the first pad. A metallic solder is disposed on an end of the post, and a second semiconductor substrate is provided having a second pad connected thereto. The length of the post is preferably from about five to about eight times greater than the thickness of the first pad. The thickness of the first pad ranges from about 4 microns to about 7 microns and the length of the post ranges from about 20 microns to about 56 microns. The fluxless method also includes depositing an unfilled polymeric liquid on the second pad; aligning and contacting the metallic solder with the unfilled polymeric liquid; and forcing by pressure the first and second semiconductor substrates toward each while simultaneously heating the metallic solder and the unfilled polymeric liquid to form a metallurgical joint between the second pad and the metallic solder.
The unfilled polymeric liquid preferably comprises, consists essentially of, or consists of from about 96% by weight to about 98% by weight of an epoxy resin and from about 2% by weight to about 4% by weight of an imidazole catalyst. Alternatively, unfilled polymeric liquid preferably comprises, consists essentially of or consists of from about 60% by weight to about 90% by weight of an epoxy resin; from about 8% by weight to about 36% by weight of a novolac resin; and from about 2% by weight to about 4% by weight of an imidazole catalyst. The unfilled polymeric liquid remains a liquid while the metallic solder is being aligned and contacted with the unfilled polymeric liquid.
The forcing by pressure of the first and second substrates toward each other preferably comprises passing the metallic solder through the unfilled polymeric liquid and contacting the second pad with the metallic solder with a pressure ranging from about 200 psi to about 500 psi, preferably from about 300 psi to about 400 psi. The heating of the metallic solder and the unfilled polymeric liquid preferably comprises increasing the temperature of the metallic solder until the temperature of the metallic solder is greater than its melting temperature; and subsequently decreasing the temperature of the metallic solder and the unfilled polymeric liquid until the temperature approximates the curing temperature of the unfilled polymeric liquid. Increasing the temperature of the metallic solder preferably comprises continually increasing the temperature for a period of time ranging from about 10 minutes to about 50 minutes; and the decreasing of the temperature of the metallic solder comprises continually decreasing the temperature for a period of time ranging from about 2 minutes to about 15 minutes until the temperature reaches around the curing temperature of the unfilled polymeric liquid where it is maintained for a period of time ranging from about 30 minutes to about 90 minutes.
In another embodiment of the present invention there is provided a fluxless method for forming an interconnect assembly comprising providing a first semiconductor substrate having connected thereto a first pad and a second pad spaced from the first pad at a distance of less than about 0.20 mm. A first post and a second post are respectively connected to the first pad and to the second pad, and a metallic solder is disposed on an end of the first post and on an end of the second post. The fluxless method also includes providing a second semiconductor substrate having connected thereto a second first pad and a second second pad. The latter pad is spaced less than about 0.20 mm from the second first pad. The fluxless method also includes depositing an unfilled polymeric liquid on the second first pad and on the second second pad. The respective metallic solders on the ends of the first and second posts are aligned with and contact the second first pad and the second second pad; and the first and second semiconductor substrates are forced by pressure toward each other while simultaneously heating the respective metallic solders and unfilled polymeric liquid to form a metallurgical joint between the respective metallic solders and the second first pad and the second second pad. In one preferred embodiment, the unfilled polymeric liquid comprises, consists essentially of, or consists of from about 96% by weight to about 98% by weight of an epoxy resin and from about 2% by weight to about 4% by weight of an imidazole catalyst. In another preferred embodiment, the unfilled polymeric liquid comprises, consists essentially of, or consists of from about 60% by weight to about 90% by weight of an epoxy resin; from about 8% by weight to about 36% by weight of a novolac resin; and from about 2% by weight to about 4% by weight of an imidazole catalyst.
Preferably, the unfilled polymeric liquid remains a liquid while the metallic solder is being aligned and contacted with the unfilled polymeric liquid. The forcing by pressure of the first and second substrates toward each other comprises passing the metallic solder through the unfilled polymeric liquid and contacting the second first pad and the second second pad with the respective metallic solders with a pressure ranging from about 200 psi to about 500 psi. As previously indicated, the heating of the metallic solder and the unfilled polymeric liquid comprises increasing the temperature of the metallic solder until the temperature of the metallic solder is greater than its melting temperature, and subsequently decreasing the temperature of the metallic solder and the unfilled polymeric liquid preferably comprises decreasing the temperature until the temperature approximates the curing temperature of the unfilled polymeric liquid. As also previously indicated, increasing the temperature preferably comprises continually increasing the temperature for a period of time ranging from about 10 minutes to about 50 minutes; and the decreasing of the temperature preferably comprises continually decreasing the temperature for a period of time ranging from about 2 minutes to about 15 minutes until the temperature reaches around the curing temperature of the unfilled polymeric liquid where it is maintained for a period of time ranging from about 30 minutes to about 90 minutes.
Embodiments of the present invention also provide an interconnect assembly comprising a first semiconductor substrate; a first pad secured to the first substrate; and a second pad secured to the first substrate and spaced from the first pad at a pitch having a value less than about 0.20 mm. The interconnect assembly also includes a semiconductor substrate having a second first pad and a second second pad secured thereto with the second second pad spaced from the second first pad at a pitch having a value less than about 0.20 mm. A first post is connected to the first pad and to the second first pad through a first metallurgical joint; and a second post is connected to the second pad and to the second second pad through a second metallurgical joint. A first cured polymeric liquid is in contact with at least a portion of the first metallurgical joint; and a second cured polymeric liquid is in contact with at least a portion of the second metallurgical joint. The first pad has a first thickness and the second pad has a second thickness. The first post has a length which is from about five to about eight times greater than the first thickness. The second post has a length which is from about five to about eight times greater than the second thickness.
These provisions together with the various ancillary provisions and features which will become apparent to those skilled in the art as the following description proceeds, are attained by the methods and assemblies of the present invention, preferred embodiments thereof being shown with reference to the accompanying drawings, by way of example only, wherein: